Create mixtral_server.py

mixtral_server.py

@@ -0,0 +1,14 @@
+from transformers import AutoModelForCausalLM, AutoTokenizer
+import torch
+
+model_name = "mistralai/Mixtral-8x7B-Instruct"
+tokenizer = AutoTokenizer.from_pretrained(model_name)
+model = AutoModelForCausalLM.from_pretrained(model_name, torch_dtype=torch.float16, device_map="auto")
+
+def generate_response(prompt):
+    inputs = tokenizer(prompt, return_tensors="pt").to("cuda")
+    outputs = model.generate(**inputs, max_length=500)
+    return tokenizer.decode(outputs[0], skip_special_tokens=True)
+
+# Testando a IA
+print(generate_response("Olá, como posso te ajudar?"))